Data-Driven Resilient Supply Management Supported by Demand Forecasting

Abstract

The article discusses several challenges related to resilient supply management and demand forecasting. Both of those topics are of great importance for food retailers and producers who aim at reducing the risk of lost sales opportunities and food waste. In the investigated case study of FitBoxY.com, due to the overestimated demand and too large deliveries, historically, even 30% of the products were overdue. The developed ML framework integrated with the supply management system enabled optimization of business costs and reduced food waste from overestimated demand. The experimental evaluation showed that, with the developed solution, it is possible to improve demand forecasting by nearly 50% compared to estimates proposed by human operators.